International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue:04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
CYBER ATTACK PREDICTION USING MACHINE LEARNING ALGORITHM Mrs.Ajitha S M 1, Abhirami B 2, Lohitha Sai S 3 , Meena M 4, Saranya K 5 1 Assistant Professor of Department of Information Technology, Meenakshi College of
Engineering, KK Nagar, Chennai. 2,3,4,5 Student of Department of Information Technology, Meenakshi College of
Engineering, KK Nagar , Chennai. ----------------------------------------------------------------------***--------------------------------------------------------------------contributing to resilient digital systems and mitigating risks posed by malicious activities. Multinomial Naive Learning (PCAML) involves leveraging diverse datasets and machine learning algorithms to forecast cyber threats. By Bayes employs probabilities for text classification, analyzing system logs, and user behavior, PCAML models, while SVM tackles both linear and non-linear data. utilizing techniques Support Vector Machines, and logistic Logistic Regression offers a straightforward approach, regression, Multinomial Naïve Bayes and TFIDF, achieve high and TF-IDF quantifies term importance based on accuracy and low false positives. This technology promises frequency, aiding in cyber threat prediction.
Abstract - Predicting Cyber Attacks with Machine
to revolutionize cybersecurity practices, offering early threat detection and proactive defense strategies. Despite challenges, such as accuracy and evolving threats, PCAML holds significant potential for enhancing organizational security and mitigating risks posed by cyber threats.
2. EXISTING SYSTEM
The existing system employs machine learning solutions to detect patterns in cyber-security data breaches. Algorithms such as logistic regression, decision trees, SVMs, and neural networks are implemented using Django, Scrapy, and Beautiful Soup.
Key Words: PCAML, Machine learning, forecast cyber attacks, TFIDF, Mitigating risks. 1. INTRODUCTION
Machine learning algorithms, including TF-IDF, Logistic Regression, SVM, and Multinomial Naive Bayes, bolster cybersecurity by analyzing historical data and patterns, accurately forecasting malicious activities. TF-IDF extracts crucial textual features, identifying key cyber threat indicators. Logistic Regression classifies attacks probabilistically, providing interpretable insights into risk factors. This multifaceted approach enhances security posture, marking a significant advancement in cyber threat prediction.
These solutions enable efficient handling of security algorithm implementation. Leveraging Django's robust framework, alongside Scrapy and Beautiful Soup for web scraping tasks, ensures comprehensive data collection. By integrating various machine learning algorithms, the system can effectively identify patterns indicative of cyber-security threats, enhancing overall detection and response capabilities against potential data breaches. The system's reliance on outdated techniques hampers its ability to adapt to evolving cyber threats. Neglecting real-time data undermines its responsiveness to immediate dangers. Scalability limitations and data management flaws impede efficient processing of large datasets. Addressing these disadvantages requires a transition to modern technologies and methodologies. Real-time data integration, scalable infrastructure, streamlined data management, and simplified architecture are essential for enhancing system effectiveness and adaptability.
1.1 ALGORITHM USAGE AND SELECTION In cyber attack prediction, Multinomial Naive Bayes utilizes probabilities for text classification. SVM handles linear and non-linear data, ideal for complex decision boundaries. Logistic Regression offers a straightforward approach to binary classification tasks. TF-IDF quantifies term importance based on frequency, aiding in cyber threat prediction. These algorithms collectively enhance cybersecurity by accurately identifying and anticipating threats, © 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 2256